“Domain Awareness” is not a new concept. Go back in history some 2,500 years, and you will find in Chapter X of the Sun Tzu clear guidance to commanders to “understand their terrain” — in today’s words: “be domain aware”.
Through historical notes, sketches, and paintings we know that savvy and successful warriors throughout history have used miniaturized battlefield models, and through “war games” have simulated variants of their armies and opponents’ engagements, seeking the best possible outcome. Generals, field officers, and intel sources slowly and painfully collaborated to gather data, and muster the mind-power needed to outsmart and neutralize an enemy threat.
Thirty years ago we were still using parallel rolling rulers and maneuvering board (MOBOARDS) papers onboard nuclear powered vessels, to execute maneuvers designed to remain in control of the “domain” (the sea) and stay ahead of our adversaries. The miniaturized battlefield model had been reduced to two-dimensional models in papers and computer screens, and we were tracking dozens of targets with limited maneuvering options, operated by crews trained in a certain style or defined operational doctrine.
Even today, with substantial “compute power” and hundreds of fully coordinated and integrated available sensors, staying “domain aware” on land, air, and sea has been challenging. Now imagine space and thousands of objects, each with anywhere from one to six degrees of freedom interacting as systems, and as systems of systems. Yes, my head hurts too; that is why we need Artificial Intelligence (AI) to truly attain Space Domain Awareness (SDA).
I was tempted to use the “checkers vs. chess” analogy, and then comparing AI Augmented SDA to “three dimensional chess," but that still misses the mark, and by a long shot.
You see, the advanced computer algorithms that we have used until now have relied on simplifications driven by the constraints of the world we experience. For example and thinking about 3D chess; that means running a stack of moves and follow-ups based upon a known initial condition (the last move), and a series of moves based on the remaining pieces and their allowed moves.
AI would be injecting factors like “what are the lighting and temperature in the room like, and what was their effect on player “A” last 1000 times this move was made?” “What hour, day, and month of the year is the match taking place, and what was the effect on the combination of A&B players before?” So on and so forth, through machine learning and other practices like process harvesting, AI SDA is much more.
AI SDA is not just about a super computer running multiple algorithms on the data streams of hundreds of sensors and expediting the work dozens of analysts tracking satellites and debris in space. Not either just predicting a conjunction in space based on the current telemetry and known capabilities of any one satellite relative to another object. Beyond advanced predictions, AI SDA is farther beyond, and closer to interpreting the possible, and injecting “knowledge” into the picture. Knowledge that a small army of analyst could not attain in a lifetime.
Imagine the capability to identify a 0.02%/year variance in the eccentricity of a satellite and being able to determine that it is due to a pinhole leak on a fuel valve that has failed on another class of satellite.
Imagine the capability to identify hostile intentions that will take place a year from now based on simple orbital maneuvers of two other satellites, controlled by seemingly independent controllers today.
If the above sounds like a stretch, I’ll offer that over the last fifty years we’ve gone from landing on the moon, to landing on asteroid Bennu, a piece of rock 200 million miles away, in the middle of space, 5 year after launch. Oh, and OSIRIS-REx is returning to earth in 2023 with a sample from it.
Space is a domain that is much more complex, and less forgiving, than any other we’ve tried to master to date. Relying on Kepler physics and super computing is only going to get us so far in achieving true SDA, especially when so many different entities are vying operational advantage or tactical superiority. While we can predict the motion of objects in orbit, and that of other RSO’s already identified, we are not fully capable of interpreting the impact of business or national interest actions taken today, in Earth or space, on tomorrow’s SDA picture.
That is why AI is so essential to SDA - we won’t get it by physics alone.
In a recent webinar to industry, the newly standing Space Force pleaded to industry to be “Space Smart”. I fully agree, and to get there we need Space Domain Knowledge, which is what you get when you marry SDA and AI.
Komentar